In the field of wireless communication in recent years, in many systems, MIMO (Multiple Input Multiple Output) transmission has been adopted that utilizes a plurality of transmit and receive antennas and is capable of performing high-speed transmission without widening a frequency bandwidth. In general, in the MIMO transmission, since a plurality of data streams are transmitted using the same frequency, MIMO signal detection becomes necessary in a receiving device.
In a MIMO signal detection technique, as an optimum detection technique, maximum likelihood detection (MLD) has been used. This is a technique for detecting transmission signal candidates maximizing a likelihood function, from among all transmission signal candidates. Since the number of transmission signal candidates exponentially increase with the number of constellations or the number of transmission streams of a modulation method, in the MLD there has been a problem that a calculation amount becomes quite large.
In NPL 1, a technique has been described in which the calculation amount of the MLD is reduced by efficiently reducing the number of transmission signal candidates. In NPL 1, by searching for a transmission signal candidate in the direction of noise enhancement serving as a factor causing an error to occur in MMSE (Minimum Mean Square Error) detection serving as linear detection in which a calculation amount is very small while a transmission characteristic is not good, a calculation amount is significantly reduced from that of the MLD while performance is prevented from being lowered.